AWS Machine Learning Specialty Exam Preparation Guide

If you searched for how to prepare for the AWS Certified Machine Learning – Specialty (MLS-C01), the essential fact comes first: you can no longer take this exam. AWS retired MLS-C01 on 31 March 2026, and as of August 2026 it can be neither sat nor renewed. Existing holders keep an active certification for three years from the date they earned it. Any study plan, course or question bank still promising to get you through MLS-C01 is out of date.

That does not mean your preparation effort is wasted — far from it. The skills MLS-C01 examined, a blend of machine learning (ML) theory and hands-on Amazon SageMaker implementation, now live in AWS's replacement credentials. This guide redirects your prep: what the retirement means for you specifically, which current exam to target, and how to adapt an ML-Specialty-style study plan to the exams that actually exist. For the historical record of what MLS-C01 covered — its domains, format and prerequisites — see the AWS Machine Learning Specialty certification guide; this article stays focused on preparation decisions.

What the retirement means depending on where you stand

You were mid-preparation for MLS-C01

Your ML fundamentals work transfers almost entirely; only the exam target changes. AWS's announcement of an expanded AI portfolio positioned two effective successors at the levels most MLS candidates came from — the AWS Certified Machine Learning Engineer – Associate (MLA) and the AWS Certified AI Practitioner (AIF) — plus, at Professional level, the new AWS Certified Generative AI Developer (AIP-C01). Pick your new target from the table below, then carry your notes across.

You already hold the ML Specialty

Your certification remains valid for three years from when you earned it, but it cannot be renewed by retake. Before it lapses, plan a transition exam — most holders will find MLA the natural continuation, and while your certification is active you qualify for AWS's 50% discount voucher on the next exam, which meaningfully cuts the cost of transitioning.

You wanted "the hardest AWS ML credential" for your CV

The specialty tier no longer has an ML entry. The most advanced current AWS credential in the AI space is the Generative AI Developer – Professional (AIP-C01), launched in beta in November 2025 with standard registration open as of March 2026, priced at $300 USD with the usual three-year validity. It validates integrating foundation models into production applications rather than classical ML breadth, so check its exam guide before assuming it matches your goals.

Choosing your replacement exam

Factor	AI Practitioner (AIF-C01)	ML Engineer – Associate (MLA-C01)	Generative AI Developer – Professional (AIP-C01)
Level	Foundational	Associate	Professional
Best for	People who work with AI/ML solutions rather than building them	Practitioners building and operating ML on AWS — closest fit for former MLS-C01 candidates	Engineers putting foundation models into production
Format	65 questions, 90 minutes; includes new ordering, matching and case-study question types	65 questions, 130 minutes; also includes the new question types	See official exam guide (format details not restated here)
Cost (USD, varies by country)	$100	$150	$300
Passing score (scaled)	700	720	Confirm in the official exam guide
Recommended background	Familiarity with AI/ML on AWS; no building required	1+ year with SageMaker and related services	2+ years cloud, 1+ year hands-on generative AI
Validity / renewal	3 years	3 years	3 years

No prerequisites exist for any of these — AWS experience recommendations are guidance, not gates. For most readers who were preparing for the Machine Learning Specialty, MLA is the answer: it is the exam that tests the MLS-style mix of ML lifecycle knowledge and SageMaker implementation at a current, bookable credential.

One timing caveat matters if you choose MLA. An update is in progress: registration for MLA-C02 opens on 1 September 2026, and the last day to take MLA-C01 in English is 28 September 2026 (the Korean, Japanese and Simplified Chinese versions of MLA-C01 continue until MLA-C02 reaches those languages). What changes in MLA-C02's content has not been published in detail, so do not rely on second-hand claims — if you are exam-ready now, book MLA-C01 before its English cut-off; if you are months away, plan against the MLA-C02 exam guide once AWS publishes it.

Adapting an ML Specialty study plan to today's exams

The MLS-C01 preparation formula — theory plus SageMaker practice plus timed questioning — still works. Here is how each strand carries over.

Keep: ML theory fundamentals

The conceptual backbone you built for MLS-C01 — framing problems as ML problems, preparing data, selecting and evaluating models, recognising overfitting and knowing what to do about it — remains the substance current AWS ML exams reward. Continue studying theory in a decision-oriented way: not "what is this technique?" but "given this data and this business goal, which approach and why?" That is how scenario questions have always been framed, and nothing about the portfolio change altered it.

Keep, and deepen: SageMaker hands-on work

AWS recommends a year or more with SageMaker and related services for MLA candidates, which tells you where the exam's centre of mass sits. Work through the full lifecycle in a sandbox account: getting data in, training, tuning, deploying, and monitoring a model in production. Practitioners whose experience is notebook-only should give deployment and operations disproportionate time — implementation-focused exams punish candidates who have trained models but never run one behind an endpoint.

Change: question-format practice

Here the retirement genuinely changes your prep. MLS-C01 was classic multiple choice and multiple response; AIF-C01 and MLA-C01 were the first AWS exams to add ordering, matching and case-study question types. Ordering questions ask you to sequence steps; matching pairs items; case studies hang several questions off one scenario. None of these reward different knowledge, but they punish unfamiliarity under time pressure, so make sure some of your practice exposes you to the current formats rather than recycled MLS-style items only. Legacy AWS Machine Learning Specialty practice questions retain value for drilling the underlying ML reasoning — ExamPractice offers free samples with fuller sets for subscribers — provided you treat them as concept practice for a retired blueprint, analyse why each answer is right rather than memorising it, and pair them with the official exam guide of the exam you will actually sit.

Change: your scope document

Throw out the MLS-C01 exam guide as a planning tool and download the current guide for your chosen exam from AWS's certification pages. Map your existing notes onto its task statements, keep what still appears, and flag what is new. Ten minutes with the official guide saves weeks of preparing for the wrong blueprint — the exact failure mode this retirement created.